Automatic pruning and lightweight deployment design of discharge depth diagnosis model based on reinforcement learning
1
State Grid Qinghai Provincial Electric Power Company Ultra High Voltage Company, Xining 810000, China
Submission date: 2025-07-24
Final revision date: 2025-12-02
Acceptance date: 2025-12-09
Online publication date: 2025-12-13
Publication date: 2025-12-13
Corresponding author
Xianghao Ding
State Grid Qinghai Provincial Electric Power Company Ultra High Voltage Company, Xining 810000
State Grid Qinghai Provincial Electric Power Company Ultra High Voltage Company, Xining 810000
KEYWORDS
TOPICS
ABSTRACT
This article proposes a lightweight model automatic pruning method based on reinforcement learning, and designs a model lightweight deployment scheme. On the server side, deep reinforcement learning is used for intelligent agent training, and automatic search is performed by interacting with the original discharge diagnosis model to determine the pruning rate of each layer; Then, the geometric median based filter pruning (FPGM) method is used to distinguish the importance of the filter and implement parameter pruning. The simulation experiment results show that this method achieves over 85% parameter compression effect on the lightweight series models MobileNetV1 and V2, as well as the classical series neural network ResNet50. The compressed lightweight model was converted into lightweight ONNX format, saved on a portable computer, and implanted into the Raspberry Pi Pico intelligent terminal through wireless transmission, achieving fault experiment simulation of substation equipment discharge on the intelligent terminal. The test results show that the discharge diagnosis model deployed after pruning using this method has significantly improved performance indicators such as memory usage, power consumption, and inference time.
FUNDING
This study was supported by the State Grid Qinghai Electric Power Company Science and Technology Project Funding (522821250004).
REFERENCES (28)
1.
Wu M, Jiang W, Luo Y. Multisource partial discharge pattern recognition in GIS based on improved SSD. High Voltage Engineering. 2023;49(2):812-821.
2.
Luo G, Zhang D, Tseng KJ, He J. Impulsive noise reduction for transient earth voltage-based partial discharge using wavelet-entropy. IET Science, Measurement and Technology. 2016;10(1):69-76. https://doi.org/10.1049/iet-sm....
3.
Firuz K, Vakilia M, Phun BT, Blackburn TR. Partial discharges pattern recognition of transformer defect model by LBP & HOG features. IEEE Transactions on Power Delivery. 2019;34(2):542-550. https://doi.org/10.1109/TPWRD.....
4.
Hussein R, Shaba KB, El-Hag AH. Robust feature extraction and classification of acoustic partial discharge signals corrupted with noise. IEEE Transactions on Instrumentation and Measurement. 2017;66(3):405-413. https://doi.org/10.1109/TIM.20....
5.
Raymond WJK, Illia HA, Bakar AHA, Mokhlis H. Partial dis charge classifications: review of recent progress. Measurement. 2015;68:164-181. https://doi.org/10.1016/j.meas....
6.
Lecun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521(7553):436-444. https://doi.org/10.1038/nature....
7.
Duan L, Hu J, Zhao G, Chen K, He J, Wang SX. Identification of partial discharge defects based on deep learning method. IEEE Transactions on Pow er Delivery. 2019;34(4):1557-1568. https://doi.org/10.1109/TPWRD.....
8.
Gao J, Zhu Y, Zheng Y. Pattern recognition of partial discharge based on VMD-WVD and SSAE. Proceedings of the CSEE. 2019;39(14):4118-4128.
9.
Davari N, Akbarizadeh G, Mashhour E. Corona detection and power equipment classification based on GoogleNet-AlexNet: An accurate and intelligent defect detection model based on deep learning for power distribution lines, IEEE Transactions on Power Delivery. 2022;37(4):2766-2774. https://doi.org/10.1109/TPWRD.....
10.
Choi S, Chung S, Lee S, Han S, Kang T, Seo J. TB-ResNet: bridging the gap from TDNN to resNet in automatic speaker verification with temporal-bottleneck eenhancement. In: 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). 2024;10291-10295. https://doi.org/10.1109/ICASSP....
11.
Wang Y, Yan J, Sun Q, Li J, Yanget Z. MobileNets convolutional neural network for GIS partial discharge pattern recognition in the ubiquitous power internet of things context: optimization, comparison, and application. IEEE Access. 2019;7:150226-150236. https://doi.org/10.1109/ACCESS....
12.
Yang T, Chen Y, Sze V. Designing energy-efficient convolutional neural networks using energy-aware pruning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017:6071-6079. https://doi.org/10.48550/arXiv....
13.
Molchanov P, Tyree S, Karras T, Aila T, Kautz J. Pruning convolutional neural networks for resource efficient inference. arXiv: 1611.06440, 2017. https://doi.org/10.48550/arXiv....
14.
Huang Z, Wang N. Data-driven sparse structure selection for deep neural networks. In Proceedings of the 15th European Conference on Computer Vision. 2018;317-334. https://doi.org/10.1007/978-3-....
15.
Lei S, Hu H, Chen B, Tang P, Tian J, Qiu X. An array position refinement algorithm for pencil beam pattern synthesis with high-order Taylor expansion. IEEE Antennas and Wireless Propagation Letters. 2019;18(9):1766-1770. https://doi.org/10.1109/LAWP.2....
16.
Tang M, Liu N; Yang T, Fang H, Lin Q, Tan Y. Free prune: An automatic pruning framework across various granularities based on training-free evaluation. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems. 2024;43(11):4033-4044. https://doi.org/10.1109/TCAD.2....
17.
Kumar A, Vashishtha G, Gandhi CP, Zhou Y, Glowacz A, Xiang J. College of Mechanical and Electrical Engineering, Wenzhou University, Wenzhou, China. Novel convolutional neural network (NCNN) for the diagnosis of bearing defects in rotary machinery. IEEE Transactions on Instrumentation and Measurement. 2021;70:1-10. https://doi.org/10.1109/TIM.20....
18.
Kumar A, Vashishtha G, Gandhi CP, Tang H, Xiang J. Tacho-less sparse CNN to detect defects in rotor-bearing systems at varying speed. Engineering Applications of Artificial Intelligence. 2021;104:104401. https://doi.org/10.1016/j.enga....
19.
Kumar A, Gandhi CP, Zhou Y, Vashishtha G, Kumar R, Xiang J. Improved CNN for the diagnosis of engine defects of 2-wheeler vehicle using wavelet synchro-squeezed transform (WSST). Knowledge-Based Systems. 2020;208:10645. https://doi.org/10.1016/j.knos....
20.
Kumar A, Zhou Y, Gandhi CP, Kumar R, Xiang J. Bearing defect size assessment using wavelet transform based Deep Convolutional Neural Network (DCNN). Alexandria Engineering Journal. 2020;59(2):999-1012. https://doi.org/10.1016/j.aej.....
21.
Kumar A, Gandhi CP, Zhou Y, Kumar R, Xiang J. Latest developments in gear defect diagnosis and prognosis: A review. Measurement. 2020;158:107735. https://doi.org/10.1016/j.meas....
22.
Kumar A, Gandhi CP, Tang H, Sun W, Xianget J. Latest innovations in the field of condition-based maintenance of rotatory machinery: A review. Measurement Science and Technology. 2024;35(2):022003. https://doi.org/10.1088/1361-6....
23.
He Y, Liu P, Wang Z, Hu Z, Yang Y. Filter pruning via geometric median for deep convolutional neural networks acceleration. In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2019:4335-4344. https://doi.org/10.1109/CVPR.2....
24.
Lin WF, Tasi DY, Tang L, Hsieh CT, Chou CY, Chang PH. ONNC: A compilation framework connecting ONNX to proprietary deep learning accelerators. In: 2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS). 2019:214-218. https://doi.org/10.1109/AICAS.....
25.
Liang Y, Ren Y, Yang X, Zha W. 3D data scattergram image classification based protection for transmission line connecting BESS using depth-Wise separable convolution based CNN. Journal of Modern Power Systems and Clean Energy. 2025; 13(2):609-621. https://doi.org/10.35833/MPCE.....
26.
Wang D, Hu M. Deep deterministic policy gradient with compatible critic network. IEEE Transactions on Neural Networks and Learning Systems. 2023;34(8):4332-4344. https://doi.org/10.1109/TNNLS.....
27.
Chen Y, Peng X, Wang H, Zhou J, Zhang Y, Liang Z. Generator stator partial discharge pattern recognition based on PRPD-grabcut and DSC-GoogleNet deep learning. IEEE Transactions on Dielectrics and Electrical Insulation. 2023;30(5): 2267-2276. https://doi.org/10.1109/TDEI.2....
28.
Hanis D, Pallotta L, Hadj-Rabah K, Bouaraba A, Belhadj-Aissa A. Covariance Matrix Estimation via Geometric Median in Highly Heterogeneous PolSAR Images. IEEE Geoscience and Remote Sensing Letters. 2025;22:2502805. https://doi.org/10.1109/LGRS.2....
Share
RELATED ARTICLE
| eISSN: | 2449-5220 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
